EchoSight

EchoSight is a conversational AI vision assistant designed to help blind and visually impaired users better understand their surroundings through natural voice interaction. EchoSight combines computer vision, depth estimation, spatial reasoning, navigation logic, and real-time voice AI into one system. A camera image is analyzed using YOLO object detection and Depth Anything V2 to identify objects, estimate their relative depth, determine their position, and understand whether they may obstruct the user’s walking path. The spatial reasoning layer converts raw computer vision results into meaningful information such as “a person to your left,” “a car directly ahead,” or “an obstacle nearby.” The navigation engine then evaluates the scene and produces actionable guidance such as “Continue,” “Caution,” “Move left,” “Move right,” or “Stop.” AssemblyAI powers the conversational voice interface, allowing users to interact with EchoSight naturally instead of relying on a visual interface. Users can ask questions about what is around them and receive spoken responses based on the current scene. The goal is to demonstrate how modern AI agents can connect perception with reasoning and voice interaction to create more accessible technology. EchoSight is currently a prototype and is not intended to replace certified mobility or safety equipment, but it demonstrates a foundation for an intelligent, conversational navigation assistant.